{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Project1-v2

# Project1-v2 - gCSCI 561 Foundations of Artificial...

This preview shows pages 1–4. Sign up to view the full content.

gCSCI 561 Foundations of Artificial Intelligence Fall 2010 Instructor: Sofus A. Macskassy Project 1: A* Search (100 points) Due: October 11, 2010 1. Introduction In this project, you are required to use C/C++ or JAVA as the programming language to solve a navigation problem of a game character in a grid world. You will be implementing the A* and Adaptive A* algorithm from scratch, and analyzing the behavior of the algorithm. 2. Project Description Consider grid-worlds with square cells. Cells are either blocked or unblocked. The start cell of the game character is unblocked. The game character does not know which ones of the other cells are blocked. However, it always senses which of its four adjacent cells is blocked and remembers this information for later use. The game character can only move from its current cell to one of the four adjacent cells at each step, the cost of move to blocked cells is infinity (which means the game character cannot move across the blocked cells), for the unblocked cell, the cost is 1. Its objective is to move from its start cell to a given goal cell with the minimum cost. It does this by always moving on a shortest presumed path (measured in steps) from its current cell to the goal cell. A presumed path is a path that requires the fewest costs for the game character to reach the goal cell. It stops when it reaches the goal. Note that the game character may need to re- plan immediately after it has observed its current presumed shortest path is blocked. The following figure shows an example of 5*5 grid terrain with black grids referring to the blocked grids, white grids referring to the unblocked grids. Here, Start cell = Agent (A). Goal cell = Target (T).

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
In the following figure , the three grids with “X” in the middle are the gri ds that Agent(A) could sense in the starting state. The other shaded ones are the grids that A cannot sense. Although there is another blocked grid in the center of the terrain, the agent cannot observe that in the first step, so the current presumed shortest path for A is shown as in the following figure. As A moves along the estimated shortest paths, the area explored by A is increased. (the explored areas are shown by white grids) But when A makes the third move, it could identify the third blocked grids, and its pre- estimated shortest path should be changed to avoid the newly observed blocked grid. So the agent needs to search again to re-plan a new path based on the current information available. The process could be shown as the following figures.
So far, we have introduced the background of the problem with unblocked cells with cost = 1. But in the project, you need to face more complicated grids with unblock cells maintaining different costs. The following picture is an example of such grids: The number inside each cell indicates the cost to reach that cell. The goal of the agent is still to reach the target. Other settings are the same as above. For the grids are unexplored, the agent could assume it as unblocked cell with cost 1.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}